On the topic of face recognition, I recall a heartwarming tale of parental love and guidance in the New York Times. It concerned a Tuesday evening back in October 2018 when a billionaire retailer, John Catsimatidis, was having dinner at an upscale Italian restaurant in Manhattan’s SoHo neighborhood, when his daughter, Andrea, walked in with her date, a young man unknown to the protective father. The billionaire did what any loving father would do in the same situation, which was to ask the waiter to go and take a photograph of her beau, which he then uploaded to the facial recognition application, Clearview AI, on his phone. He was immediately presented with collection of photos of the man along with their source and was able to determine immediately that is daughter’s escort was a venture capitalist from San Francisco, at which point he sent his daughter the full biography of her companion by text.
(I would have assumed that at this point he would take the standard set of actions, which is to call for security and have this poor choice of life partner ejected. But no. Rather amusingly Mr. Catsimatidis, who presumably has never met a VC before, commented that “I wanted to make sure he wasn’t a charlatan”.)
Well, what self-respecting father could do more. But… are we ready for face recognition everywhere?
The start-up behind the Clearview app has a database of billions of photos, scraped from sites such as Facebook, Twitter and LinkedIn. Even one of its backers says that it “might lead to a dystopian future or something”. Well, at least a dystopian future in which face recognition databases are used to stop deadly viruses from spreading has some merits and, indeed, some people might well choose that option. But it’s the “or something” that bothers me. The unknown unknowns, so to speak.
In a 2023 interview, the CEO of Clearview, Hoan Ton-That, said that more than 2,400 police agencies across the US are customers and that the fees of $2,000 per annum for each officer with access to the system is “pretty inexpensive, compared to what’s come previously”. Police use of the technology is a live issue here in the UK where there have been many recent newspaper stories about face recognition because of these decision by the police to begin using it despite the current not-quite-Minority-Report state of the art. The facial recognition technology used by police in London incorrectly identified members of the public in 96 per cent of matches made between 2016 and 2018. As my good friend Jamie Bartlett, the man behind “The Missing Cryptoqueen” (the best podcast of 2019), observed on this police use of face recognition in London “if the technology doesn’t work it will be a disaster and if it does work it will be even worse”.
My point, though, is we don’t really know what the new etiquette should be, which is why some people (eg, the EU, Google and me) would like to see a moratorium in the deployment of face recognition technology (a moratorium that ought to be extended to all population-scale passive biometrics, in my opinion). As the FT noted, Brussels and Silicon Valley rarely see eye-to-eye when it comes to technology regulation but in this particular case (and in the case of artificial intelligence) they may well be aligned. We don’t quite know what we are doing.
This is a valid concern. China illustrates quite clearly how quickly the technology can begin to penetrate. I can choose any one of a thousand examples to illustrate this point, but I like this one: taxi drivers in the Chinese city of Xi’an are verified by facial recognition technology when they get behind the wheel. The biometric identification system is, as is much the fashion these days, linked to an AI to ensure that drivers are not misbehaving (eg, using their smartphone when on the road and so forth). Now, I can see why such a system is attractive. Who doesn’t want a safer taxi service?
(Of course, if we start to rely on such interfaces, they can bring unexpected problems. I particularly enjoyed this story from the South China Morning Post about a woman who had plastic surgery only to discover she could no longer pay online or get into her office!)
We cannot hold back the tide and there is no need to
Which ever way you look at it, Regulators are surely right to focus down on face recognition as being a technology with a social context that we absolutely do not understand.
By the way, one area that doesn’t get anything like as much attention as it should is the issue of face recognition but for animals. This, as the Wall St. Journal noted, is a challenge because “it’s not like you can tell a donkey to stand still“. Quite. However I can tell you that it can be done. I was privileged to have a speaker from JD Digits, a subsidiary of JD (China’s largest e-commerce business) on my panel about AI ethics and governance at the Innovate Finance Global Summit (IFGS) last year. JD Digits, amongst other things, runs face recognition services for farmyard animals such as cows and pigs. It turns out that pig face recognition is a big business. There are 700m pigs in China and the productivity gains that farmers can obtain from ensuring that each pig is fed optimally, that sick pigs are kept away from the herd (and so on) are very significant.
(Apparently the face recognition system also goes some way to reigning in wannabe Napoleons, as Dr. X explained that there are some “bully pigs” that try to obtain a disproportionate share of barnyard resources. The system can spot them chowing down when they shouldn’t be and flag for intervention.)
Snippings 